SAN FRANCISCO, CA — YCombinator alum Athelas, a deep learning based biotech company, is unveiling a rapid blood diagnostics and immune monitoring platform that can be used at home by chemotherapy patients, as well as in oncology research by Pharma companies. The company has also closed $3.7 million in funding led by Sequoia Capital, with participation from Initialized Capital and Joe Montana’s Liquid2. Angel investors include Color Genomics co-founder Elad Gil, James Hong, Stanford Deep Learning Professor and Salesforce Chief Scientist Richard Socher, and former Y Combinator COO Qasar Younis. Advisors and research affiliation include Dr. Sam Gambhir, Chair of Stanford Radiology, pioneering Neutropenia researcher Dr. David Dale, and Dr. Preet Choudhary, Chief of Hematology & Blood Diseases at USC Keck School of Medicine.

The Athelas technology uses deep learning and computer vision to analyze high resolution blood images to generate cell counts. The Athelas image analysis backend is particularly well suited to overcome many of the problems traditionally associated with drop based diagnostics — such as inaccurate counting due to debris and variations in particle size & morphology — allowing consumers to receive end-to-end, lab grade results at home within 2 minutes. This clinically validated technology has implications in oncology, particularly for cancer patients and oncologists looking to customize chemotherapy treatments and dosing based on an individual’s recovery metrics, allowing for better outcomes.

“Athelas is bringing cancer patients a quick and reliable way to test their blood levels from within their home,” said Alfred Lin, partner at Sequoia. “Their new platform empowers patients to confidently monitor their condition and will cut down on unnecessary urgent care visits. We believe in Tanay and Deepika’s bold vision to transform at-home blood tests into an easy and accurate diagnostics tool that’s as trusted as a thermometer.”

Clinically Validated Tests

Founded by former Stanford researchers Tanay Tandon and Deepika Bodapati, who have backgrounds in computer vision, machine learning and molecular imaging, Athelas’ tests have been clinically validated at FEMAP hospital — a high volume ER/physician clinic, with published accuracies against Labcorp venous counts (considered the gold standard), and in bench studies at Stanford University. End-to-end usability testing with untrained patients operating the Athelas system found a correlation coefficient of r=0.96 compared to results from venous bloods draws at LabCorp (Figure 1). Third party clinical studies led by Myriam Navarro and Dr. Gustavo Martinez at FEMAP hospital found fingerpick samples on the Athelas system matching venous blood draws across various CBC parameters on the Sysmex 5000 with n=0 clinical range errors across the 76-person (aged 9–86) study cohort (Figure 2).

Figure 1. User Generated Results Compared to Gold Standard. More information available in Athelas Paper, “Rigorous Bench Studies to Determine Sensitivity & Precision of Athelas Strip & Device At-Home Usability and Interpretability Studies for Consumers”

Figure 2. Graph of third party validation of Athelas WBC results compared with WBC results from FEMAP Sysmex 5000. More information available in Athelas Paper, “Clinical Third Party Accuracy Validation of the Athelas Device CBC Device Compared with Venous Sysmex 5000 in POC Clinic”

The key behind this accuracy is Athelas’ proprietary test strip combined with the latest breakthroughs in deep learning and computer vision. The system directly images the cells themselves, instead of examining flow cytometry or impedance — techniques that only focus on the sizes of particles rather than their entire morphological profile. As such, Athelas’s use of visual deep learning to tag cells enables higher accuracies, precisions and a wider range of tests on fingerstick samples than previously possible.

“The Athelas rapid blood diagnostics device has revamped patient care at FEMAP hospitals by allowing the doctors to diagnose patients more precisely within a matter of seconds, facilitating immune monitoring at the patient’s bedside, and granting our population an extensive set of low cost tests,” said Myriam Navarro of FEMAP. “In one case, Athelas enabled FEMAP to diagnose a leukemia case with their handheld device and a drop of blood that otherwise went undetected by doctors.”

New Partnerships

With the new funding, Athelas will be partnering with oncologists and pharmaceutical companies to rapidly test immune markers at patient homes, and to develop custom diagnostics on its deep learning platform. Athelas aims to help the hundreds of thousands of cancer patients proactively monitor febrile neutropenia — a potentially deadly complication caused by treatment.

Tandon and Bodapati started the company after approaching the problem in-lab from two different sides — Tandon’s background in visual machine learning to tag morphologies, and Bodapati’s in rapid staining and biological imaging. Tandon was previously an engineer at Wit.ai, an API for building voice-activated interfaces that was acquired by Facebook, and deep learning startup MetaMind acquired by Salesforce. Bodapati was previously a researcher at the Stanford Multimodality Imaging lab with her work on imaging brain cancer cells published in more than four publications, including Science Translational. They recently hired Dhruv Parthasarathy, who was director of machine learning at Udacity, to lead the engineering team.